Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Raise error for long prompt #273

Merged
merged 7 commits into from
Jul 1, 2023
Merged

Conversation

LiuXiaoxuanPKU
Copy link
Collaborator

This is a fix for #113. The program hangs when the input prompt is too long because the check will always succeed and the request will always in the waiting queue and will never be added to the running queue.
Add a check before the if statement, if the prompt length is too long, it will raise ValueError directly.

@zhuohan123
Copy link
Member

Hi Lily, please check all the possible situations here and make sure you can handle all of them. Thanks!

@LiuXiaoxuanPKU
Copy link
Collaborator Author

I move forward with the length check. The original code should already handle the len(prompt) + len(generated) > limit case, so I only deal with model limit and len(prompt) > limit. Let me know if there are any problems, thanks!

Copy link
Member

@zhuohan123 zhuohan123 left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Hey Lily, thanks for the work! Please check my comments.

vllm/core/scheduler.py Outdated Show resolved Hide resolved
vllm/sequence.py Outdated Show resolved Hide resolved
vllm/sequence.py Outdated Show resolved Hide resolved
vllm/sequence.py Outdated Show resolved Hide resolved
vllm/engine/arg_utils.py Outdated Show resolved Hide resolved
vllm/config.py Outdated Show resolved Hide resolved
@zhuohan123 zhuohan123 mentioned this pull request Jun 30, 2023
Copy link
Member

@zhuohan123 zhuohan123 left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks for the hard work! In general LGTM! Will merge it after the following small comments are fixed.

vllm/engine/llm_engine.py Outdated Show resolved Hide resolved
vllm/config.py Outdated Show resolved Hide resolved
@zhuohan123 zhuohan123 merged commit dafd924 into vllm-project:main Jul 1, 2023
michaelfeil pushed a commit to michaelfeil/vllm that referenced this pull request Jul 1, 2023
@LiuXiaoxuanPKU LiuXiaoxuanPKU deleted the fix branch July 6, 2023 17:55
hongxiayang pushed a commit to hongxiayang/vllm that referenced this pull request Feb 13, 2024
sjchoi1 pushed a commit to casys-kaist-internal/vllm that referenced this pull request May 7, 2024
yukavio pushed a commit to yukavio/vllm that referenced this pull request Jul 3, 2024
…project#273)

SUMMARY:
* update benchmarking, testing, and accuracy jobs to run on label
`aws-test-a10g-24G` or `aws-test-a10-96G` which is based on "vanilla
deeplearning" AMI
* update relevant GHA actions and workflows to not be dependent on
`pyenv` virtualenv
* update "model cache" to use local disk as opposed to "EFS"

TEST PLAN:
runs on remote push

---------

Co-authored-by: andy-neuma <andy@neuralmagic.com>
Co-authored-by: Domenic Barbuzzi <domenic@neuralmagic.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants